Announcing Gloo Open

Today, we are thrilled to announce Gloo Open, our new open source artificial intelligence initiative, signaling a crucial step toward transparency and collaboration in AI. Gloo Open reflects our commitment to offer values-aligned AI for human flourishing through our research, models, non-copyrighted datasets, and standards framework. This launch, supported by industry leaders and partners, showcases our dedication to shaping a future where AI development is guided by shared values and open collaboration.

“AI is so critical to the future of humanity — the principles of open and trust are essential,” said Pat Gelsinger, Gloo Chairman and former CEO of Intel. “At Gloo, we are committed to participating fully according to the best practices of the open source community. Working together we will ensure AI innovation is accelerated and social trust is developed through an open approach.”

At the heart of Gloo Open is our innovative approach to creating values-aligned AI models for our users—including developers, publishers, and parachurch organizations. We begin by cloning established open source models (such as those from Meta and Mistral) onto our secure, isolated infrastructure. From there, our proprietary training pipeline fine-tunes these models using values-aligned data sets. This allows each user or organization to control the values their model will uphold, helping minimize censorship issues and biases found in many mainstream AI models.

“At Gloo, we believe that open ecosystems are far more powerful than closed ones,” explained Steele Billings, our Chief AI Officer. “The Gloo AI fine-tuning process, developed through our early Flourishing AI research, leverages an open approach, ensuring that the outputs are aligned with the values of the user and culturally attuned.”

We are also leveraging findings from the Global Flourishing Study to develop advanced AI models for human flourishing. These findings, produced in a five-year research collaboration involving the Human Flourishing Program at Harvard, Baylor’s Institute for Studies of Religion, Gallup, and the Center for Open Science, will be released for open source use later in 2025.

Our ongoing commitment to open source AI mirrors similar efforts by industry giants like Meta, Hugging Face, and others. To further elevate our mission, we will introduce a comprehensive framework of standards for evaluating AI models on values-driven criteria. Our framework not only measures performance metrics, but also alignment with core values that may include community engagement, family life, spiritual formation, vocational health, and physical or mental wellness. It addresses accuracy to specific content such as the Bible, as well as historical and theological contexts—ensuring that AI outputs consistently reflect the ethos of their developers and users.

“I am incredibly proud of the work our team has dedicated over the years to bring Gloo Open to life,” emphasized Scott Beck, Gloo CEO. “Thanks to the relentless dedication of our brilliant team and the valued partners who share our commitment to open approaches and trust, we’re now ready to share our research and models with the global developer community. Our commitment to innovation and transparency remains unwavering, and Gloo Open marks a significant step forward in shaping a future of values-aligned AI that enriches lives and drives human flourishing.”

We invite developers interested in accessing Gloo Open or building on our Gloo AI platform to pre-register at ai.gloo.com/open.